FreeChat is a native macOS application designed to provide users with a seamless, offline experience for interacting with Large Language Models (LLMs). It targets both AI enthusiasts and general users unfamiliar with LLMs, offering a private, local alternative to cloud-based AI services like OpenAI's ChatGPT, with a focus on simplicity and performance.

How It Works

FreeChat leverages the llama.cpp library to run various gguf formatted LLM models entirely on the user's local machine. This approach eliminates the need for internet connectivity for core functionality and ensures all conversations are saved locally, prioritizing user privacy and data security. The application is built using Swift, aiming for a native macOS feel and performance.

Quick Start & Requirements

• Installation: Available via TestFlight beta, Mac App Store, or by building from source using Xcode.
• Prerequisites: macOS, Xcode for source builds. Compatible with any gguf formatted model supported by llama.cpp.
• Resources: No specific hardware requirements beyond standard macOS compatibility are listed, but performance will depend on the chosen LLM and the Mac's hardware.
• Links:
  • TestFlight Beta: https://6032904148827.gumroad.com/l/freechat-beta
  • Mac App Store: https://apps.apple.com/us/app/freechat/id6458534902
  • Source Build: Open mac/FreeChat.xcodeproj in Xcode.

Highlighted Details

• Runs LLMs locally on macOS without internet connection.
• Compatible with any gguf formatted model from sources like Hugging Face.
• System prompts are customizable to change model personas and expertise.
• Conversations are saved locally and privately.

Maintenance & Community

The project acknowledges contributions from Georgi Gerganov (llama.cpp), Meta (Llama 2), Jon Durbin (Spicyboros model), TheBloke (model quantization), and Monica Kogler (logo/UX). There is no explicit mention of community channels like Discord or Slack in the README.

Licensing & Compatibility

The README does not explicitly state a license. Given its reliance on llama.cpp (MIT License) and Meta's Llama models (custom license), users should verify compatibility for commercial or closed-source use.

Limitations & Caveats

The project is described as prioritizing simplicity and performance over features. Advanced settings like prompt format and temperature are planned but not yet implemented. A "Personas" feature for saving system prompts and model settings is also on the roadmap.